Geomapping Drive-Time Based Market Areas for DoD TRICARE Beneficiaries Richard R. Bannick, Ph.D. and Tyler Erickson, Ph. D.* Accompanied in the audience by James Laramie* & Amii Kress Oct 25, 2006 Office of the Assistant Secretary of Defense (Health Affairs)/TRICARE Management Activity- Health Program Analysis and Evaluation Note: The views expressed are those of the authors and do not reflect the official policy or position of the Department of Defense or the U.S. Government * Supported by Altarum Institute, Ann Arbor, MI
Overview Preamble: the Military Health System: spatial dispersion and geographic markets Motivation- Policy Changes Problem Statement Methodology Overview Develop Tool for Drive-time Based Market Areas Geocode Facilities Create Market Areas for All Facilities Create Market Area to ZIP Code Mapping (90% solution) Assess Market Area Population Summaries Methodology Example Results Summary- Conclusions & Next Steps Questions 2
Military Health System (MHS) Strategic Priorities Forces are medically ready and protected Medical capability can be deployed anywhere The MHS builds healthy communities 3
Military Health System (MHS) 9.2 million beneficiaries 3 Peacetime benefits options: Prime (HMO) Standard (Indemnity) Extra (PPO) 2 components Direct Care (MTF) 70 inpatient facilities 411 ambulatory clinics Purchased Care (Civilian Network) Institutions Providers Retail Pharmacies 4 Mail Order Pharmacy
TRICARE: U.S. Military Treatment Facilities & Eligible Population (Oct 2006) 5
Motivation: Invoking Policy Change TRICARE Policy for Access to Care (2006) Prime enrollee Access to Care standards: Emergency Care: where-ever available Urgent (Acute Care): appointed to visit within 24 hours and within 30 minutes travel time of the beneficiary s residence. Or offer referral and authorization care outside the civilian network Routine Care: 7 calendar days and 30 minutes travel time Wellness and Health Promotion Services: within 4 weeks and within 30 minutes travel time Referrals for Specialty Care Services: within 4 weeks and within one hour s travel time Local MTFs will use three or more publicly available web-based mapping programs to determine eligibility for enrollment or continued enrollment. 6
Motivation- Policy Implications Need to understand, and plan for, how the recent ASD (HA) policy establishing TRICARE Prime access standards might effect: MTF enrollment behavior Within Direct Care- Between MTFs Between Direct and Purchased Care Overall Program Costs Beneficiary expectations and satisfaction Efficient MTF utilization Need to respond to Congressional oversight of DoD TRICARE performance How well is the Direct Care system complying with access standards? 7
Problem Statement Current MHS distance-based market areas methods were developed in the late 1970s before the maturation of GIS and do not support these planning and reporting functions Not consistent with drive-time standards Produce distorted views in multi-service markets and highly urbanized areas Feasible developing and maintain new drive-time market areas for the over 300 Military Treatment Facilities (MTFs) in the US required an automated tool How to map entirety of the MHS corporate data warehouse into drive-time markets geocoding it is not practical Cost & time prohibitive, complicated by continual population movements Significant Personal Health Information (PHI) and security obstacles Can 90% solution be developed that is: Efficient and cost-effective to maintain and implement in MHS information systems Creates no added PHI or security burden 8
Comparative Results Example Current and Potential Markets 9
Methodology Overview Develop Tool for Automating Creation of Drive-time Based Market Areas Geocode Facilities Create 30 Minute Drive-time Market Areas for All Facilities Create Market Area to ZIP Code Mapping (90% solution) Assess Fit of 90% Solution ZIP Code based Population Data Geocoded Population Data 10
Methodology (Continued) Develop Tool for Automating Creation of Drive-time Based Market Areas Geocode Facilities Create 30 Minute Drive-time Market Areas for All Facilities Create Market Area to ZIP Code Mapping (90% solution) Assess Fit of 90% Solution ZIP Code based Beneficiary Data Geocoded Beneficiary Data 11
Drive-Time Market Area Creation Tool The Drive-time maximum (in minutes) represented by the perimeter of the drive-time polygon Multiplier used to adjust the drivespeeds to account for local traffic conditions (e.g., reduction for rush hour) The road network dataset that describes the interconnected road segments in terms of length, road class, and travel rates ESRI shapefile with geocoordinates of MTFs; drive-time travel calculated independently for each MTF coordinate Maximum distance allowed between the facility (point) and the nearest road segment (line). The drive-time calculations are based on nearest road segment, rather than the actual point. The folder in which the calculated line network and polygon shape files are saved. One polygon and one line shapefile are created for each facility. 12
Running the Drive-Time Tool Toolbox Method 13
Network Analyst Extension Issues Analysis of large quantities of facilities causes memory faults Solved by using Python scripting and using two geoprocessor objects StreetMap Pro network dataset cannot be called from Python script (bug related to licensing) 14
Methodology (Continued) Develop Tool for Automating Creation of Drive-time Based Market Areas Geocode Facilities Create 30 Minute Drive-time Market Areas for All Facilities Create Market Area to ZIP Code Mapping (90% solution) Assess Fit of 90% Solution ZIP Code based Population Data Geocoded Population Data 15
High Resolution Used to Geocode Facilities Evans Army Community Hospital (ACH) geocoordinates set to those in Google Earth Fire/Hospital Layer 16
Example Result of Geocoding Evans ACH location actually three Miles NNW from flight line at Ft. Carson 17
Methodology (Continued) Develop Tool for Automating Creation of Drive-time Based Market Areas Geocode Facilities Create 30 Minute Drive-time Market Areas for All Facilities Create Market Area to ZIP Code Mapping (90% solution) Assess Fit of 90% Solution ZIP Code based Population Data Geocoded Population Data 18
Drive Time Network NH Bremerton 19
Drive Time Network USAF Academy 20
Methodology (Continued) Develop Tool for Automating Creation of Drive-time Based Market Areas Geocode Facilities Create 30 Minute Drive-time Market Areas for All Facilities Create Market Area to ZIP Code Mapping (90% solution) Assess Fit of 90% Solution ZIP Code based Population Data Geocoded Population Data 21
Create Market Area to ZIP Code Mappings the 90% Solution Estimated baseline populations in drive-time market areas for evaluating the suitability of alternative methods of ZIP Code assignment to drive-time market areas Baseline Population Estimate Population counts weighted by the percent of the ZIP Code area contained in the drive-time (e.g., 20% of ZIP Code area in drive-time market area, assign 20% of population in ZIP Code to drive-time market area) METHOD Evaluated three methods for assigning ZIP Codes to drive-time market areas by comparing their resulting population counts to the baseline population Low estimate High estimate Centroid estimate DEFINITION Include ZIP Code if it is completely contained in drive-time area Include ZIP Code if any portion is contained in drive-time area Include ZIP Code if its geographic centroid is contained in drive-time area 22
Methodology Example 30-minute Drive-Time - USAF Academy Baseline Estimate* Low Estimate High Estimate Centroid-based Estimate Too few ZIP Codes Too many ZIP Codes Best representation of market area Population estimate = 58,177 Population estimate = 34,544 Population estimate = 79,102 Population estimate = 57,774 * Validation with geocoded beneficiary data in progress 23
Methodology (Continued) Develop Tool for Automating Creation of Drive-time Based Market Areas Geocode Facilities Create 30 Minute Drive-time Market Areas for All Facilities Create Market Area to ZIP Code Mapping (90% solution) Assess Fit of 90% Solution ZIP Code based Population Data Geocoded Population Data 24
350,000 Results Fit of 90% Solution The centroid-based method is highly correlated with the baseline population for the area 300,000 Centroid- Based Population Population 250,000 200,000 150,000 100,000 Pearson s r = 0.99616 50,000 - - 50,000 100,000 150,000 200,000 250,000 300,000 350,000 Baseline Population Centroid Estimate 25 1:1 ratio line
Summary & Next Steps GIS Network Analyst approach provides useful capability for deriving macrolevel drive-time market areas Provides ability to measure & manage MHS compliance with access policy Easy to implement in MHS information systems Is efficient and not costly to implement and maintain May be maintained monthly with existing MHS data operation Creates no additional PHI or security burden to existing MHS operations Next Steps: Assess error of baseline estimate with geocoded beneficiary data and reassess fit of 90% solution Develop gold standard for MTF geocoordinates Test sensitivity of drive-time boundary files (e.g., posted speed vice rush hour); decide MTF specific driving assumptions Refine and finalize data & methodology, including outliers Apply methodology to analysis of other access standards (e.g., trauma center access standards) 26
Thank You! Questions? Richard Bannick, Ph.D. Director, Performance Evaluations OASD(HA)/TMA- Health Programs Analysis & Evaluation 703-681-3636 x5035 richard.bannick@tma.osd.mil Amii Kress, MPH Policy Analyst OASD(HA)/TMA-HPA&E 703-681-3636 X 5032 amii.kress@tma.osd.mil Tyler Erickson, Ph.D. Senior Analyst Altarum Institute 734-302-4794 tyler.erickson@altarum.org James Laramie Senior Healthcare Analyst Altarum Institute 734-302-4634 27
BACK UP SLIDES- NOT FOR PRESENTATION BUT FOR DISCUSSION 28
So How Can GIS Help Us? Related GIS distance-based Access Studies Euclidean/Straight line (for paired data, from centroid of population geographic shape, e.g. census block, to item of interest: Lin (2004): Access to Community Pharmacies by the Elderly in Illinois: A Geographic Information Systems Analysis. J. Medical Systems, 28(3): 301-309. St. Claire, et al. (1981). Technical Annex, MHSS Catchment Area Directory, Vector Research, Incorporated and Actuarial Research Corporation, DHA-1 WP81-1. Road Networks: Branas, et al. (2005). Access to Trauma Centers in the United States, JAMA, 293(21): 2626-2633. 29